I'd need to learn more about the C implementation of connection deduplication before deciding whether it makes sense for grpc-go. E.g.s: Is it optional or always enabled? Does it scale connections when the maximum streams limit is reached?
Could you file a bug against grpc-go for this if you think it's worth investigating further? Thanks, Doug On Friday, October 26, 2018 at 2:19:57 PM UTC-7, Ruslan Nigmatullin wrote: > > Hi, > > It looks like grpc-core performs a deduplication of TCP connections > for subchannels with the same channel arguments [1]. However it > doesn't look like it's a case for grpc-go implementation. > It sounds like a useful behavior in the context of grpclb, as it > allows reducing number of connections to the remote balancer (as > multiple channels can point to the same remote balancer server). > > Is it just a missing functionality at grpc-go side or some fundamental > difference in implementations? > > Thanks > > 1. > https://github.com/grpc/grpc/blob/a0578f075ccd64a7041c02c0efcf3bb590aa019e/src/core/ext/filters/client_channel/subchannel_index.cc > > -- You received this message because you are subscribed to the Google Groups "grpc.io" group. To unsubscribe from this group and stop receiving emails from it, send an email to grpc-io+unsubscr...@googlegroups.com. To post to this group, send email to grpc-io@googlegroups.com. Visit this group at https://groups.google.com/group/grpc-io. To view this discussion on the web visit https://groups.google.com/d/msgid/grpc-io/3f8511ee-8cbb-4e23-9e36-cdfcda5f4725%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
